The Case against Fair Value Accounting - Securities and Exchange
Transcript of The Case against Fair Value Accounting - Securities and Exchange
1
The Case against Fair Value Accounting*
By A. Rashad Abdel-khalik
University of Illinois
Mailing address:
College of Business
2037 BIF
515 E. Gregory
Champaign, IL 61820
*1 This paper is based on a lecture delivered at the CGA Accounting Conference, University of Ottawa,
Canada on October 31, 2008 and at the Chinese Accounting Forum held at Nanjing University on
December 6, 2008.
2
The Case against Fair Value Accounting
Synopsis
A brief discussion of the current views on fair value accounting is followed by reviewing the
concerns raised by the critics. These concerns are related to (a) reliability and verifiability of the
underlying assumptions in the case of mark to model, and (b) the recognition of large losses
when prices decline. My concerns about fair value accounting rest on three different issues: (1)
the potential for redistribution of capital as dividends when cash dividends exceed realized
earnings; (2) reporting failure as success, and success as a failure when increasing credit risk of
debtors allows them to recognize gains and vice versa, and (3) increasing information asymmetry
when combining realized and unrealized gains and losses as well as failing to disclose the drivers
of holding gains and losses. These problems with fair value accounting do not necessarily mean
that we should abandon fair value accounting. Instead, the proposal is made to produce two sets
of financial statements: one based on historical cost and the other based on fair value. Each type
serves a different purpose. Mixing and blending historical costs and fair values is problematic in
that it worsens transparency and enhances information asymmetry.
The Issues
Fair Value (FV) accounting was enthusiastically sponsored by some leading academics
since the early 1960s1 but it was given a push back in those days by policy makers who seemed
to favor conservative valuation that is largely based on harder (documented) evidence. In recent
1 See, for example, Edwards and Bell on Entry Value (1961), and Chambers (1966) and Sterling (1970) on Exit
Value.
3
years, especially with the 1998 adoption of Statement of Financial Accounting Standard (SFAS)
# 133 on accounting for derivatives, FV has taken a new and a totally different life that was
manifested in the promulgation of SFAS # 157 (2006) on measurement bases and SFAS # 159
(2007) on giving the management of reporting firms the option to adopt fair value in reporting
financial assets and liabilities that have not thus far been included in the standards that call for
FV. The train of FV took on more supporters with the anticipated change from Generally
Accepted Accounting Principles (GAAP) to International Financial Reporting Standards (IFRS).
In fact, accelerating the planned convergence of GAAP to IFRS was cited in the FASB’s minutes
as the motivation for adopting SFAS # 159.
It has been argued by most ardent supporters of FV that reporting either quoted or
estimated market values of assets and liabilities is a reflection of ―truth.‖ However, greater
doubts2 about the relevance and reliability of FV have come to surface mostly after the onset of
the liquidity crisis and the current recession. For a set of different reasons, I question any
relevance of the income statement or even the usefulness of net income as a diagnostic measure
of management performance if FV accounting continues to be implemented as charted by either
the Financial Accounting Standards Board (FASB) or the International Accounting Standards
Board (IASB).
Fair Value Accounting and the Economy
The linkage between accounting and the economic system is transparent to many
participants in the information environment; most observers know that the efficient functioning
2 See for example, David Katz, ―Fair-Value Revolution,‖ CFO Magazine, September 1, 2008.
4
of capital markets is conditional on the availability of relatively timely and reliable information.
Yet, there are those who see the linkage only in a crisis mode, as is the case at the present
environment. The deepening of the sub-prime liquidity crisis will definitely mark a watershed in
the history of accounting. Many outspoken influential individuals in business and government
have not been reluctant to reveal their strong-held, though mistaken, belief that the adoption of
fair value (mark-to market) accounting had, in no small way, been a contributing factor in
causing the crisis. Being so convinced, they continue to call for the repeal of the FV accounting,
a call that merits consideration if standard setters insist on requiring a single set of financial
statements, a point that will be elaborated at the conclusion of this commentary.
While considering the use of fair value per se is not new in the field, the issue has
recently come to the spotlight for two reasons: (i) the promulgation of SFAS #157 (2006); and
(2) the need to recognize the sharp decline in market values as losses.
SFAS # 157 provides guidance on the measurement of FV. Three levels or stages are
offered. At Level I, FV is based on quoted market prices; at Level II, FV is based on observable
inputs when active markets do not exist; and in the absence of observable inputs, preparers make
a choice from among many estimates, which is Level III. It is that third level, which is
sometimes referred to as mark-to-model that has generated most of the criticism against fair
value accounting. In particular, for Level III estimation, preparers would have to make
―reasonable‖ assumptions about expected future cash flow amounts, timing and uncertainty and
about the relevant cost of capital to be used in discounting projected cash flows to present. The
estimated present values are then used as the FV estimates at which assets and liabilities are to be
5
reported. Changes in FV estimates for financial assets and liabilities are to be recognized in the
income statement.
Since the beginning of the economic downturn, mark-to-model has been under attack
primarily due to the questioned validity of the invoked assumptions and the reliability of the
estimated FV measures and the open suspicion of management manipulation. The question
arises as to whether mark-to-model is actually a reasonable estimate of today’s market price of
an asset for which there is no current market. This question became more critical for large
portfolios of bank loans and mortgages that were either Trading Securities or Available for Sale
but became illiquid by movements in external market conditions. In fact, the entire philosophy
of FV accounting was, I believe, seriously undermined when banks were allowed to reclassify
such loans to the ―held-to-maturity‖ category so that they would not recognize the changes in
market values. For example, Citigroup was allowed to reclassify about $65 billion dollars to
held-to-maturity.3
The argument presented by the opponents of fair value accounting is straightforward.
Simply stated, in weak economic conditions, FV accounting leads to generating pessimistic
assumptions that precipitate significant decrease in asset values and significant lowering of
earnings because of recognizing the unrealized losses in the income statement. FV accounting
and especially mark-to-model , argue the critics, (i) is subjective and arbitrary; (ii) reflects poorly
3 See, for example, ContiGroup Annual Report (2008, pp. 88)
In millions of dollars Carrying value at December 31, 2008
Debt securities reclassified from Trading account assets to held-to-maturity investment $33,258 Debt securities reclassified from available-for-sale investments to held-to-maturity investments 27,005 Total debt securities reclassified to held-to-maturity investments $60,263 Debt and equity securities reclassified from Trading account assets to available-for-sale investments $ 4,654
6
on the performance of the company with no evidence to back it up; (iii) has the potential of
significantly increasing the volatility of reported earnings; and (iv) has the potential of
misleading investors into projecting the worsening financial conditions into the future.
According to the critics of fair value accounting, these four conditions gave reasons to
investors to dump their shareholdings during the year 2008 and to cause the equivalent of a run-
on-the-bank reminiscent of the 1930s. Yet, on the upside, when prices were on the rise, none of
the opponents of FV had any concerns or second thoughts about taking unrealized gains to
income. Is seeking fair value on the upside and throwing it away on the downside a sign of
schizophrenic preference? Some would argue that such is the case.
Does Fair Value Accounting have Bad Consequences?
As an academic, I am obliged to ask the question: Is it possible that the opponents of fair
value accounting are justified? Does the use of fair value at any measurement level, mark-to-
market, or mark-to-model have undesirable consequences? My response to this question is in
the affirmative but largely for reasons different from those offered by the critics of FV
accounting. While the critics have concentrated on the reliability and the ―hardness‖ of the
measures of FV, my basic claim is that posting changes in market values to the income statement
will have three major negative consequences. These are outlined below and are discussed
further:
1. Potential redistribution of capital as dividends due to confusing ―earned
income‖ with ―expected or holding capital gains and losses.‖
2. Reporting ―failure as ―success and success as failure.
7
3. Increasing information asymmetry and reducing transparency.
1. The Potential for Redistributing Capital as Dividends.
Under SFAS #157 and SFAS #159 (as well as the earlier SFAS #133 on derivatives and,
to some extent, SFAS #123R on share-based compensation), changes in FV find their way to the
income statement. Even in the case of cash flow hedge, the changes in fair values that have been
accumulated in Other Comprehensive Income are to be transferred to the income statement when
the forecasted transactions are firmed up. Irrespective of the approach or the measurement level
used, whether market price quotes or estimates based on applying some valuation models, it is
clear that FV measures are the outcome of neither arms-length exchanges nor the transfer of risk
of ownership to entities outside the firm. Fair values are the expected prices at which the firm
may be able to sell the assets in the normal course of business if they were to be sold today.
Instead, fair values are only expectations the realization of which is conditional on many
factors—primarily management decision to liquidate the position and market stability or
volatility. To this extent, this conditionality may permit labeling changes in FV as hypothetical.
From a different perspective, changes in FV are essentially holding capital gains (or losses)
that have not been earned (realized) as of the time of financial reporting. It must be emphasized
further that the conditionality of changes in expectations implies an underlying probability
distribution with the eventual realization of some change might be considered highly likely
(probable) while other changes have only a remote probability of being realized. Reporting these
changes (expected holding gains and losses) in the income statement is in effect giving the same
weights to a highly probable outcome as the weight given to an outcome with a remote
8
probability of occurrence, and, in fact, giving both of these conditional outcomes the same
weight as that given to a sure, realized, outcome.
Moreover, when the realized and unrealized gains and losses are aggregated with earned
income without distinction, a more serious problem arises because of the potential for
redistribution of capital as dividends. To make this point salient, consider three periods in a
company’s life during which all reported earnings were distributed as dividends. :
9
Reported Period 1 Period 2 Period 3
Total
Activity
Capital 1,000.00 1,000.00 1,000.00
Retained Earnings (beginning) 0 5.00 11.00
Net Income:
20.00
17.00
(11.00)
26.00
Cash Dividends 15.00 11.00 0 26.00
Retained Earnings (End of period) 5.00 11.00 0
Not Reported
Components of Net Income:
Realized
Unrealized gains (losses)
10.00
10.00
7.00
10.00
2.00
(13.00)
19.00
7.00
Cumulative Excess Dividends over
Realized Earnings
5.00 9.00 7.00 7.00
In this case, implementation of FV accounting led to including both realized and unrealized
holding gains (and losses) in reported earnings. Cash dividends absorbed all reported earnings,
which exceeds the amount of realized earnings by as much as the total unrealized holding gains.
The failure to disaggregate the realized and unrealized gains has led, in this case, to
redistributing capital. In other words, the excess of cash dividends over realized earnings is
nothing other than re-distribution of capital. Admittedly, the illustration is simplified enough to
make the point that the firm’s management has no constraint on distributing what we,
accountants, tell them to be the measurement of ―net income‖ under GAAP.
When I presented this drawback of FV accounting at different conferences, there was always
someone who notes that it is the management’s responsibility not to distribute cash dividends
beyond realized earnings. Unfortunately, we do not even care to explicitly inform the
management of how much of the reported earnings is realized; there is no evidence that the
disaggregation of net income into realized and unrealized components has been a concern. Nor
has it been given serious consideration, as far as I can tell, by standard setters, or even
10
academicians. It is very likely that it will take a redesign of the accounting system of any
company to permit generating the types of data that would inform the management of such
breakdown between realized and unrealized components.
2. Reporting Failure as a Success and Success as a Failure
This concern relates to the valuation of financial liabilities, whether standalone instruments
or derivatives. To highlight this particular drawback, I will focus on fixed-income instruments.
Valuing such instruments at their fair values and taking the resulting changes to the income
statement give rise to this phenomenon of reporting failure as success, and vice versa, as
elaborated below.
In general, the value of fixed-income financial instrument moves opposite of the movement
of changes in market rates of interest (cost of capital). As cost of capital increases, investors
demand higher return than the fixed-income rate paid by the instrument. Consequently, the
market (fair) value of the instrument decreases, and vice versa. Under FV accounting, the
borrowers who issue fixed-income instruments will recognize gains equal to the decline in the
market value of their debt that arises from an increase in interest rates and will recognize losses
as the market value of their debt increases with the decline in the market interest rates. In an
economic sense, the fair value of debt is the real obligation conditional on (a) liquidating the
position, and (b) market stability.
Even when the market rate of interest is stationary and there is an active market, an
unfortunate outcome of FV accounting is that the same outcome noted above would be obtained
as the borrower’s creditworthiness changes. This is precisely where FV accounting leads to
11
nonsensical and perverse results. With a decline in credit worthiness, borrowers become more
risky as their risk of default increases. To compensate for the increased risk, capital market
participants will demand higher risk premium and the market value of debt declines.
Conversely, when creditworthiness improves and the borrowers have superior credit risk, the
required rate of return on their debt declines because the superior credit rating means that market
participants will require a lower risk premium than is being charged. The increase in the value of
debt is a loss to the borrower and, under FV accounting; such loss would be recognized in the
income statement.
The two above-noted scenarios create a dilemma of different dimensions. On the one hand,
both scenarios have the same outcome in that lower market values result in the debtors’
recognition of gains, while higher market values of debt results in the debtors’ recognition of
losses. On the other hand, the two scenarios signal different information: in the first scenario,
holding gain or loss arises from external forces, while in the second scenario, holding gain or
loss emanates from internal forces. The distinction between internal and external forces is quite
relevant to anyone interested in the financial well being of the reporting firm. The suggested
high degree of relevance of these scenarios arises from the fact that the ―internal forces‖ are
pertinent to the evaluation of management performance and success. The perverse nature of the
outcome of this accounting process is that the debtors recognize gains when they become more
risky (failure is reported as success), and recognize losses when they become less risky (success
is reported as failure).
To be sure, I am not the first one to address this dilemma; in fact the minutes of the
FASB’s meetings show a discussion of this problem at different times. Expressing the prevailing
12
thought around the FASB, though not speaking for this standard setting organization, Crooch and
Upton (2001) have in fact addressed this problem and offered examples by comparing changes in
creditworthiness of two hypothetical companies. However, to this day, I am unable to
understand or be convinced of the proposition that FV accounting in this situation (i.e., reporting
gains for posing greater credit risk and reporting losses for posing lower credit risk) enhances
transparency or information relevance. To the contrary, I believe it takes away from relevance
and transparency and it could be misleading to the average investor who relies on accounting
numbers as reported to measure firm performance.
The other side, of course, is that every debt is owned by someone else outside the debtor’s
entity. As an asset for creditors, the market value of investment in fixed-income securities will
generate gains and losses opposite that of the debtors’ position. That is, the creditor will
recognize losses with a decline in the creditworthiness of the issuer (debtor), and will recognize
gains with the increase in the creditworthiness of the issuer. In the absence of evidence on
impairment of the value of the asset, the holding gains or losses are conditional expectations, just
the same as they are for the debtors. Therefore, the creditor’s recognition of these holding gains
or losses would also be misleading to average investor, especially since these changes are not
indicators of managerial performance or action. This is critical because no information is
provided on the degree and extent (or likelihood) of realizing the holding gains and losses that
are being recognized. As in the case of the debtors’ reporting, creditors’ reporting of unrealized
capital gains and losses would be aggregated with realized earnings and, once again, relevance
and transparency are brought into question.
13
Whether it is the case of debtors or creditors, the increase in market interest rate has the
same outcome as the downgrading of the debtors’ creditworthiness (increasing the debtors’ risk)
and vice versa. Two concerns with this outcome:
(1) External users of financial statements have no way of disentangling the
information to identify the source of the gain and to be able to differentiate between
the external and internal sources of the reported gains—i.e., whether it is increased
market interest rate vs. increased credit risk of the issuer.
(2) A more disturbing feature of the above analysis is that it is not really as a
hypothetical situation as the illustration above implies. Indeed, Katz (2008) reports
that the 25 firms of the S & P 500 that have ―the biggest amounts of liability
measured at fair value, widening credit spread—an indication of a lack of
creditworthiness—spawned first quarter earnings gains ranging from $11 million to
$3.5 billion.‖
3. Increasing Information Asymmetry and Reducing Transparency
Financial reporting is simply the process of transferring information about the firm from the
management’s private domain to the public domain. This process results in reducing information
asymmetry between insiders and outsiders. The blueprint of the system implicitly embodies two
principles: (a) the amount, frequency and type of information transferred to the public domain
should be sufficient to allow for proper evaluation of the firm’s performance without revealing
the industry secrets that may compromise the reporting firms’ competitive advantages; and (b)
the process must allow all external users equal access to information.
14
Given these principles, it is fair to state that the main goal of financial reporting system is to
reduce information asymmetry. It is, therefore, disconcerting to find out that FV accounting
does indeed increase information asymmetry rather than reducing it. In light of the above
discussion, two reasons arise for making this claim:
(a) It is noted above that the changes in market values of assets and liabilities held by the
firm are merely changes in expectations—i.e., subject to hypothetical or conditional
realization. It is also noted that FV accounting calls for taking these changes in
expectations for many financial instruments to the income statement. Furthermore, SFAS
# 159 allows the management the discretion to choose extending FV accounting to other
financial assets and liabilities.
It should also be recalled that the realized gains or losses are posted to the income
statement. Therefore, under FV accounting, changes in expected values are given equal
weight and reliability as the realized changes in assets and liabilities that arise from arms-
length exchange between the firm and external entities. Furthermore, not only that
unrealized holding gains and losses are reported as earnings without being subject to the
test of realization, they are aggregated with realized gains and losses in such a way that
external investors could not indentify how much of the reported earnings has actually
been earned and how much remains realizable—i.e., potential earnings conditional on the
occurrence of future events that might or might not lead to realization. Heckman (2001)
voices this concern as follows: ―The public seeking such information will look in vain in
the financial statements and must either inspect the books directly or rely on the service
of one of the rating agencies‖ (P. 72).
15
Furthermore, investors and external users of financial statements are not likely to
be able to assess the degree of future realizability of the yet-to-be-earned components.
Insiders who might be in the position of evaluating the likelihood of realization do not
provide the information that users need to make this assessment. This concern, however,
may not be as serious in some cases such as in the case of inventories with high turnover
rates—e.g., Trading Securities when there is a liquid and active market. In such cases,
the probability of realizing the changes in market values is expected to be high simply
because the duration between valuing and liquating a position is quite short, leaving low
opportunities for altering the uncertainty about realizability of market price changes.
Unfortunately, as noted in the case of Citigroup above, firms are allowed to opt away
from FV by reclassifying trading securities to a category not subject to the rules of
pricing at market.
(b) Information asymmetry increases also due to the absence of disclosing the reasons for
unrealized holding gains or losses on debt instruments. It must be noted that the
asymmetry of concern here goes beyond mixing and aggregating the realized and
unrealized gains and losses; it deals with aggregating unrealized gains and losses on debt
that masks the true underlying problems related to credit risk. When we fail to inform
investors of how much an inferior credit risk allows a debtor to recognize gains and how
much a superior credit debt allows the debtor to recognize losses, not only that we could
be accused of adopting a flawed model, but we also do not the level of transparency
needed portray reality without the need to seek the help of rating agencies as Heckman
16
notes (2001). Once again, I fail to see how this lack of transparency could not be viewed
as providing more relevant information as advocates of FV accounting claim.
In summary, keeping the investor uninformed about the source of such gains or
losses does (i) increase information asymmetry, (ii) reduces transparency, and (iii) could
potentially mislead investors.
What are the Alternatives?
The traditional view of stewardship emanated from the need of investors to know what
managers do with the capital funds entrusted to them. This is the perspective that gave rise to the
role of historical-cost basis in financial reporting because the price that a manager had paid for
an asset, coupled with documented evidence, is an indication of specific uses of funds. Funds
not used remain in a liquid (cash) account and the investors will, therefore, be able to account for
the uses of all the funds they provided to the company. Hence, the concept of stewardship has
traditionally been backward looking because it aims at providing information to answer the
question of ―what did you, the management, do with my money?‖
In contrast, fair value accounting is a forward-looking basis of reporting because the market
value of an asset (whether observed or estimated) is the discounted net cash inflows expected to
be generated from using the asset. However, by reporting the fair value alone, the investor will
have an idea of what is likely to come but will not know what happened in the past. This shift in
focus is predicated on the ill conceived notion that historical cost is an irrelevant sunk cost due to
the fact that ―relevance‖ has been given a limited interpretation. The argument goes that fair
value is more relevant because what others are willing to pay for the asset (market price) is a
surrogate for the cash flow that managers expect to generate for the investor. Thus, historical
17
cost is the actual cost of acquiring the asset (the cash given up to acquire the asset); fair value is a
measure of the opportunity cost (the cash that could be received to give up the asset)
It is not a matter of dispute that investors make decisions based on expectations and, to that
extent, fair value is in concert with investors’ outlook. However, by reporting fair values alone,
investors are deprived from the firm-specific benchmark, which is represented by historical cost
or the funds invested in acquiring the asset. Without knowledge of historical cost, the investor
would not know whether the fair value (present value of future cash flows) is better or worse the
funds given up to acquire the asset. From the investors’ standpoint, neither historical cost nor
fair value perspective offers the necessary and sufficient information to know the answers to two
concerns:
a) what the management did with their funds, and
b) what the management expects to get for that use of those funds.
The first concern calls for historical cost (the traditional stewardship perspective), while the
second concern calls for fair value (the prospective stewardship perspective). I believe that both
types of financial reports need to be produced separately, not blended! The difficulty with the
contemporary philosophy of standard setting by both the IASB and FASB, as I see it, is the
fixation on a single perspective for financial reporting at the exclusion of the other.
Dramatization of the difference between current GAAP and the proposed two-financial
statements system could perhaps be portrayed by comparing Figures One and Two below.
18
Figure One: Current GAAP
Assets Liabilities
Historical Cost XXX (XXX)
Fair Value ZZZ (ZZZ)
Mixed Attributes: Current Set
of Financial Statements
19
Figure Two: The Proposed Two Sets of Financial Statements
and their objectives.
Assets Liabilities
Historical Cost
First set of Financial
Statements
XXX (XXX)
Fair Value
Second set of
Financial Statements
ZZZ (ZZZ)
What did you do with my money?
What will you get me for my
money?
20
Concluding Comments
In this essay, I wanted to bring us back to reality—the term ―fair‖ must be stripped of any
judgment on ―fairness‖ in fair values. As a method of valuation, fair values have their place in
accounting, but as a method of income recognition, we need to step back and ask about the
unexpected consequences of granting equal weights to holding gains and losses as those given to
earned income. Changes in fair values may or may not be realized (earned) in the immediate
future or the long term. Income determination and use has always been predicated on the
assumption that net earnings are a measurement of the firm’s performance. Under FV
accounting as being currently suggested (a single set of financial statements); net income
measures both (a) firms performance and (b) effect of changing market conditions. There is
nothing wrong with these two components—investors need to know both types of information,
unless they are mixed up and blended in an ad hoc combination that conveys information about
neither.
It is noted above that investors seek information or answers (from the management) for
two questions: (a) what did you do with my money? And (b) what will you get me for my money?
To answer the first question, we need historical-cost basis (classical notion of stewardship). To
answer the second question, we need fair values (effective notion of stewardship). Investor
information needs will only be satisfied with both types of financial statements—historical basis
and fair-value basis. To insist on reporting a single set of financial statements, we will end up
with having either (a) a balance sheet consisting of mixed attributes, or (b) an income statement
consisting of mixed attributes. Neither situation addresses investors’ information needs
21
satisfactorily. It is unclear to me why we need to make a choice between purifying one of the
two important financial statements—the income statement or the balance sheet—at the expense
of the other.; we could do both if we stop focusing on having to publish only a single set of
financial reports. This position is similar to Abdel-khalik (1990) and the numerous accounting
standards that have been issued in the past two decades that have added enormous complexity to
GAAP could only enhance the merits of this proposal.
22
References
Abdel-khalik, A. Rashad (1990). The Troubles with Financial Statements: The Need for Two
sets of Financial Statements, published by Accounting Research Foundation, University
of Sydney, 1990).
Chambers, Raymond J., (1966). Accounting, Evaluation and Economic Behavior.
Englewood Cliffs, N. J. Prentice Hall.
Crooch, M G. and W. S. Upton (2001). ―Credit standing and liability measurement.‖
Understanding the Issues, FASB Report Volume 4, Series 1: pp. 1 – 6.
Edwards, Edgar O. and Philip W. Bell (1961). The Theory and Measurement of
Business Income. Berkeley and Los Angeles: University of California Press.
Financial Accounting Standards Board. Various statements at www.fasb.org
Heckman, Philip E. (2004). ―Credit standing and the fair value of liabilities: A critique,
North American Actuarial Journal, Vo. 8, No. 1 pp. 70-185
Katz, David, ―Fair-Value Revolution,‖ CFO Magazine, September 1, 2008.
Sterling, Robert S. (1970). Theory of Measurement of Enterprise Income. Lawrence,
Kansas: The University of Kansas Press.